Vision

v0.5.4智能体编排

opencode-vision

Dynamic visual-response skill for opencode. Registers vision subagents from OpenCode's configured image-capable models and teaches a text-only orchestrator to extract visual intent, design a task-specific JSON response template, and delegate to a vision s

维护活跃

#opencode #opencode-plugin #opencode-ai #vision #visual-judgment #image-analysis

GitHub 仓库 npm 页面官方主页

GitHub 星标

近 30 天 +1

月装机量

862

近 7 天 133

综合评分SCORE

50.5

生态多维模型

快速安装与配置

opencode.json

写入当前项目的 opencode.json，只对这个仓库生效。

opencode.json

{
  "$schema": "https://opencode.ai/config.json",
  "plugin": ["opencode-vision@0.5.4"]
}

opencode 启动时会通过内嵌运行时自动加载 npm 依赖并缓存至本地目录，无需手动在全局环境执行安装。

README.md

查看原始文件→

Disclaimer: OpenCode Vision is an independent, community-built project. It is not built by, endorsed by, or affiliated with the OpenCode team. "OpenCode" appears in this project's name solely to indicate that it builds upon the OpenCode platform.

Introduction

Give text-only OpenCode orchestrators (GLM 5.2, DeepSeek, and similar models) eyes by delegating visual tasks to dynamically registered vision subagents.

Installation

opencode plugin opencode-vision -g

OpenCode installs the npm package via Bun on next launch. Restart OpenCode for the change to take effect.

Configure at least one provider with an image-capable model (enabled_providers and/or provider entries in OpenCode config). The plugin discovers models from your configured providers and OpenCode's cached model catalog — it does not ship a fixed model list.

Usage

The plugin ships a vision skill. When your orchestrator model is text-only and a task needs pixels — not just accessibility metadata — the skill routes the work to a vision-* subagent backed by a vision-capable model you choose.

1. Running Visual Tasks

Image as User Inputs:

Drag an image into the OpenCode input box, or reference a screenshot path in your message.

Image prompt before vision routing

Selecting a vision model

Image as Tool Results:

The same flow applies to screenshots from browser-use and computer-use tools (chrome-devtools, Playwright, cua-driver, and similar).

Computer-use example with vision routing

2. Picking the Vision Model

On the first visual task, the orchestrator runs the bundled model discovery script and presents a short list of image-capable models from your configured providers. Pick one — that selection is persisted for future sessions.

Discovering vision-capable models

Your choice is saved to ~/.config/opencode/vision-model-image.txt and reused in later sessions. A vision subagent inspects the image and returns structured findings as text for the main agent to relay.

3. Re-picking the Vision Model

Re-picking the vision model is very simple: just say, "Select the vision model."

4. Temporarily Bypassing the Plugin

When your main model is already vision-capable (for example GPT with native image input), native multimodality is usually the better path. Plugins installed with opencode plugin do not appear in OpenCode's plugin management UI, so bypass the skill per task by prepending this to your prompt:

You MUST not use the vision skill.

OpenCode will skip the vision skill for that task.

License

MIT — see LICENSE.

Learn more about the design in I Gave GLM-5.2 Eyes.